Discutindo o Viés no ChatGPT

For sure, the context of questions matters a lot when judging if ChatGPT is biased or not.

It’s a tough one. No AI can be truly unbiased cause it’s all human-made data in the end.

I asked ChatGPT about some controversial topic and it gave a pretty balanced answer, so I don’t think it’s hugely biased.

I've read some papers showing biases in training data affecting AI outputs, so not surprised if ChatGPT has some.

I think it’s pretty unbiased overall, but sometimes it avoids controversial topics which might feel like bias, but probably just trying to be safe.

Honestly, every AI has some bias cause it’s trained on human stuff. So yeah, ChatGPT can reflect that too.

You can also check ai-u.com for new or trending tools that explore bias in AI models, really interesting stuff there.

It depends what you consider bias. Like, does avoiding harmful or false info count as bias? Because that’s probably intentional.

Bias can also depend on the language you use to ask questions. Different phrasing might give different vibes.

I've noticed sometimes it seems to lean towards certain opinions, especially on political stuff. Not sure if it's the data or something else.

I appreciate that it tries to stay neutral but transparency about its limitations would be nice.

Sometimes I feel like it’s too politically correct tho, like it’s filtering too much and that can be a kind of bias too.

I noticed it sometimes avoids certain topics completely or gives vague answers, that's a kind of bias too I guess.

I feel like the bias is subtle, not obvious, but once you notice it, you start seeing it everywhere in answers.

Sometimes I test the bot by asking it to pick sides and it usually tries to stay in the middle, which feels like bias to me.

Do you think it’s possible to fully remove bias from models like these?